Draft version April 7, 2020 Typeset using LATEX twocolumn style in AASTeX61 THE ORIGIN OF NEPTUNE'S UNUSUAL SATELLITES FROM A PLANETARY ENCOUNTER Daohai Li1 and Apostolos A. Christou2 1Lund Observatory Department of Astronomy and Theoretical Physics Lund University, Box 43, SE-221 00 Lund, Sweden 2Armagh Observatory and Planetarium College Hill, Armagh, BT61 9DG Northern Ireland, UK Submitted to AJ ABSTRACT The Neptunian satellite system is unusual, comprising Triton, a large (∼ 2700 km) moon on a close-in, circular, yet retrograde orbit, flanked by Nereid, the largest irregular satellite (∼300 km) on a highly eccentric orbit. Capture origins have been previously suggested for both moons. Here we explore an alternative in-situ formation model where the two satellites accreted in the circum-Neptunian disk and are imparted irregular and eccentric orbits by a deep planetary encounter with an ice giant (IG), like that predicted in the Nice scenario of early solar system development. We use N-body simulations of an IG approaching Neptune to 20 Neptunian radi (RNep), through a belt of circular prograde regular satellites at 10-30 RNep. We find that half of these primordial satellites remain bound to Neptune and that 0.4- 3% are scattered directly onto wide and eccentric orbits resembling that of Nereid. With better matches to the observed orbit, our model has a success rate comparable to or higher than capture of large Nereid-sized irregular satellites from heliocentric orbit. At the same time, the IG encounter injects a large primordial moon onto a retrograde orbit with specific angular momentum similar to Triton's in 0.3-3% of our runs. While less efficient than capture scenarios (Agnor & Hamilton 2006), our model does indicate that an in-situ origin for Triton is dynamically possible. We also simulate the post-encounter collisional and tidal orbital evolution of Triton analogue satellites and find they are decoupled from Nereid on timescales of ∼104 years, in agreement with Cuk´ & Gladman(2005). Keywords: Satellite formation | Neptunian satellites | Irregular satellites | Close encounters { Celestial mechanics arXiv:2004.02512v1 [astro-ph.EP] 6 Apr 2020 Corresponding author: Daohai Li [email protected], [email protected] 2 Li & Christou 180 bits on the equatorial plane of Neptune. This raises the 155 question of how they arrive at their current orbits. Harrington & Van Flandern(1979) originally postu- 130 lated an encounter between an ad hoc planetary body 50 of several earth masses and Neptune, flipping Triton's (deg) i orbit and scattering Nereid outward. This scenario has 25 been criticized (Farinella et al. 1980; McKinnon et al. 0 1995) on the grounds that the encountering planet is not observed in the solar system and that the encounter 1 may have over-excited Neptune's orbit. Also, compu- tational resources available at that time allowed only 0:1 one \successful" simulation run, making it difficult to estimate the success rate of this particular evolutionary e 0:01 path. In a subsequent model where Triton was assumed captured, Goldreich et al.(1989) suggested that Nereid 0:001 could be scattered onto a wide orbit by Triton, an out- come not reproduced in numerical simulations (Nogueira et al. 2011). 0:001 0:01 0:1 It is believed the giant planets radially migrated in a (RHill) the early solar system (Fern´andez& Ip 1984; Malhotra 1993) in the now widely-accepted framework of the Nice Figure 1. Distribution of giant planet moons in model. There, the planets formed at different heliocen- the (a; i) and (a; e) planes from Scott Sheppard's web- site: https://sites.google.com/carnegiescience.edu/ tric distances from those where they are presently ob- sheppard/moons. Symbol size is proportional to the square served and, due to interactions with a primordial plan- root of that moon's actual size. Grey symbols represent etesimal disk, they migrated to their current locations. moons around Neptune while Triton and Nereid are the large Since its introduction (Tsiganis et al. 2005), the Nice −4 and small black filled circles respectively. Small (5-10×10 ) model has evolved considerably to meet an enhanced vertical offsets have been added to allow moons with negli- set of constraints. Because of the difficulty to correctly gible orbital eccentricity to be displayed on a log-scale. excite the orbit of Jupiter (Morbidelli et al. 2009) and, at the same time, to avoid over-exciting the inner main 1. INTRODUCTION asteroid belt (Morbidelli et al. 2010; Minton & Malho- Observational biases notwithstanding, Neptune has tra 2011) and the terrestrial planets (Brasser et al. 2009; the least number of satellites among the four giant plan- Agnor & Lin 2012), Jupiter is thought to have impul- ets but perhaps with the most intriguing orbits. The sively \jumped" over the 2:1 mean motion resonance largest moon, Triton, is orbiting its host planet at 14 with Saturn, owing to close encounters with an ice giant. Neptunian radii (RNep) on a circular path but, oddly, As such, a five-planet variant of the Nice model, where in a retrograde direction. Nereid, > 200RNep further an additional ice giant planet (IG hereafter) was subse- out and the third largest moon in the system, has the quently ejected from the solar system, was introduced highest orbital eccentricity among solar system moons (Batygin et al. 2012; Nesvorn´y& Morbidelli 2012). The (Figure1). IG, before its ejection, probably encountered other plan- Mechanisms favouring capture of Triton from helio- ets as well, leading to the capture of Trojans and irreg- centric orbit include gas drag (McKinnon 1984; McKin- ular satellites (Nesvorn´yet al. 2013, 2014a) and the em- non & Leith 1995), collisions (Goldreich et al. 1989) and placement of the so-called \kernel" of the Kuiper Belt three-body gravitational interaction (Agnor & Hamilton (Nesvorn´y 2015). These planet-planet encounters may 2006; Nogueira et al. 2011; Vokrouhlick´yet al. 2008). have been as close as 0.003 au (Nesvorn´yet al. 2014b), See Colombo & Franklin(1971); Heppenheimer & Porco penetrating to the satellite region. (1977); Pollack et al.(1979); Cuk´ & Burns(2004); The appearance of the Nice scenario mitigates the two Nicholson et al.(2008); Nesvorn´yet al.(2007, 2014a) major objections (Farinella et al. 1980; McKinnon et al. for discussions on satellite capture. 1995) to the in-situ formation of Triton (Harrington & In an alternative in situ formation scenario, the two Van Flandern 1979) since the encountering IG could moons have accreted in the circum-Neptunian disk (e.g., have been ejected from the solar system and thus be ren- Szul´agyiet al. 2018) with initially circular, prograde or- dered unobservable while Neptune's eccentricity, even if Origin of Neptune's unusual satellites 3 (1) (2) imum separation denc between the IG and Neptune, we start with the Sun-IG-Neptune system at the moment Nereid Triton of the two planets' closest approach (Cloutier et al. 2015), i.e., when the relative position vector satisfies Triton encounter Neptune Neptune j~rIG−Nj = denc; we set denc =0.003 au (or 18 Neptunian IG radii Nesvorn´yet al. 2014b). At this moment, the rela- collision tive velocity vector is perpendicular to the relative posi- tion vector: ~vIG−N?~rIG−N. We further assume their rel- (3) (4) 2 ative kinetic energy j~vIG−Nj to be uniformly distributed 2 Nereid Nereid within the range (0; 3vesc) where vesc is the two-body es- cape velocity between the IG and Neptune. The orien- Triton Triton tides tations of the two vectors are random in the solid angle. Neptune Neptune Then the IG-Neptune barycentre is assigned a heliocen- tric orbit parameterised by the set of orbital elements (aIG+N; eIG+N; iIG+N), with values within the ranges: ◦ Figure 2. Illustration of our origin scenario for Triton and aIG+N 2 (10; 40) au, eIG+N ≤ 0:7 and iIG+N ≤ 10 ; Nereid. Before the encounter, several tens of pre-existing the angular elements are randomly drawn from a uni- moons orbit Neptune (1), all small (open circles) except one form distribution. We calculate the position and veloc- that is Triton-sized (black filled circle). After the IG en- counter (2), the orbit of the large moon is flipped and a ity vectors ~rIG+N and ~vIG+N and combine them with small moon is emplaced onto a wide, eccentric orbit, turning ~rIG−N and ~vIG−N to fully define the heliocentric state into Nereid. Subsequently (3), collisions with Triton remove vectors of the IG and Neptune at the instant of closest the other small moons and decouple the orbit of Triton from approach. Next, we carry out a reference frame trans- Nereid's. Finally (4), tides circularise Triton's orbit. formation such that the z-axis is parallel to the total angular momentum of the three-body system. We re- excited, may be damped owing to the interaction with fer to this as the heliocentric frame. The IG mass is 18 the planetesimal disk (actually, the encounters consid- Earth masses (Nesvorn´yet al. 2014b), similar to that of ered here ensure that the orbit of Neptune is at most Neptune. mildly excited). Hence it is worthwhile to reexamine We integrate this system backwards and forwards for the in-situ formation model within the constraints of 10 years apiece using the general Bulirsch-Stoer algo- the Nice scenario, also providing statistics of success- rithm in MERCURY (Chambers 1999) with an error tol- ful vs unsuccessful simulations runs to estimate model erance of 10−14. At the end of these two integrations, efficiency and exploring the ensuing Neptunian system we check the planets' mutual distance is larger than 0.8 evolution post-encounter.